Speech synthesis method, electronic device and storage medium

Abstract

The invention relates to the technical field of speech semantics, and provides a speech synthesis method, an electronic device and a computer storage medium, the method comprising: acquiring a presetverbal skill in a response scene, the verbal skill comprising a fixed text and a variable text; screening out timbre features corresponding to the response scene from a preset voice library accordingto the response scene to record a fixed text in the verbal skill to obtain fixed voice; performing voice synthesis on a variable text in the verbal technique according to the screened tone characteristics to obtain variable voice with the same tone characteristics as the fixed voice; and finally, splicing the fixed voice and the variable voice to generate a synthesized voice with the timbre characteristics. According to the invention, the voice with unified timbre is synthesized according to the response scene, so that the voice in human-computer interaction is natural and coherent, and the user experience is further improved.

Description

technical field [0001] The invention relates to the technical field of speech semantics, in particular to a speech synthesis method, an electronic device and a computer-readable storage medium. Background technique [0002] With the development of artificial intelligence technology, the voice broadcast in human-computer interaction requires coherence and naturalness, and the voice broadcast is repeated based on the recorded speech text. [0003] However, the existing machine does not consider the actual business scene when recording speech, and uses one tone to record various speeches in different business scenarios, so that some speech recordings do not conform to the business scene where they are located, resulting in the recorded voice tone blunt It is unnatural and the broadcast is incoherent; and the sentence transitions or intonation changes during the speech recording process can easily lead to differences in timbre before and after the recorded speech, which in turn ...

AI Technical Summary

Problems solved by technology

[0003] However, the existing machine does not consider the actual business scene when recording speech, and uses one tone to record various speeches in different business scenarios, so that some speech recordings do not conform to the business scene where they are located, resulting in the recorded voice tone blunt Unnatural, incoherent broadcasting; and sentence transitions or intonation changes during the recording of speech techniques can easily lead to differences in timbre before and after the recorded speech, which in turn affects the user experience during human-computer interaction

Images